• This study investigates effects of nitrogen and phosphorus on high Arctic heath vegetation, particularly bryophytes.
• Heath communities received factorial combinations of nitrogen (0, 10 and 50 kg ha−1 yr−1) and phosphorus (0 and 5 kg ha−1 yr−1) in five applications per growing season, for 8 yr.
• Nitrogen decreased lichen cover but did not affect cover of any other functional type. However, just 10 kg ha−1 yr−1 increased the proportion of physiologically active bryophte shoots, and decreased their nitrate assimilation capacity. Phosphorus had greater effects, and the combination of both nutrients altered species composition. Individual bryophyte species displayed contrasting responses to fertilization, suggesting that they should not be grouped as a single functional type.
• The ‘critical load’ of nitrogen for Arctic heath lies below 10 kg ha−1 yr−1. Nitrogen and phosphorus are colimiting in this sytem, so the critical load of nitrogen will be lower where phosphorus availability is greater. Responses of vegetation to any increase in net mineralisation due to soil warming will depend on the ratio in which nitrogen and phosphorus availabilities increase. The effects of nutrient enhancement are very persistent.
Arctic ecosystems are expected to be particularly sensitive to nutrient enrichment as they are strongly nutrient limited, particularly by nitrogen and phosphorus (Shaver & Chapin, 1980). Inputs of nitrogen to the Arctic have doubled in the last century directly as a result of acidic deposition (Neftel et al., 1985; Laj et al., 1992). In addition it is predicted that global warming will have its greatest impact near to the poles (Kattenburg et al., 1996). Summer surface temperatures at high latitude have already increased by 2°C over the last 30 yr (Chapman & Walsh, 1993). As soil warms, mineralisation rates are likely to increase, potentially increasing nitrogen and phosphorus availability to the vegetation, particularly if soil temperature exceeds approx. 10°C (Nadelhoffer et al., 1991, 1992; Robinson et al., 1995). It is also suggested that phosphorus availability will increase if greater thaw depth gives plant roots increased access to mineral horizons (Nadelhoffer et al., 1992). Effects of realistic increases in nitrogen and phosphorus availability, at rates that are likely to occur via these routes, are untested.
A recent report by the Arctic Monitoring and Assessment Programme (AMAP, 1997) dismissed atmospheric nitrogen deposition as not being of concern in the Arctic. However, this is not supported by any ecological evidence. Experiments have shown that even 10 kg N ha−1 yr−1 is sufficient to reduce mycorrhizal colonisation of arctic shrubs, decreasing fungal diversity (Alexander & Woodin, unpublished; Woodin, 1997). Atmospheric deposition of nitrogen in the Arctic is generally very low, approx. 1 kg N ha−1 yr−1, but in some areas already exceeds 10 kg N ha−1 yr−1, for example the Taymyr Peninsular in Russia and Northern Alaska (Woodin, 1997, based on data from Nenonen, 1991; Jaffe & Zukowski, 1993). Thus we may presume that, contrary to the AMAP statement, changes in Arctic ecosystems may already be occurring. Indeed, the predicted critical load of nitrogen for arctic heath is estimated to be within the range of 5–15 kg N ha−1 yr−1 (Hornung et al., 1995).
The Arctic is characterized by low rates of both nitrogen deposition (approx. 1–10 kg N ha−1 yr−1; see above in this section) and net mineralization (approx. 1–6 kg N ha−1 yr−1) (Nadelhoffer et al., 1992). Phosphorus deposition is less well quantified. Globally, deposition rates tend to vary between < 0.1 and 1 kg P ha−1 yr−1 (Newman, 1995) and the likelihood that deposition within the Arctic is at the lower end of that range is supported by data from subarctic Sweden (0.01 kg P ha−1 during the growing season) (Malmer & Nihlgård, 1980). Net mineralisation of phosphorus in Arctic soils is also slow and can even be negative due to microbial immobilisation (e.g. −0.25 to +0.25 kg P ha−1 yr−1) (Schmidt et al., 1999). The effects of increased soil temperature on net mineralization rates, and resultant changes in nutrient availability to vegetation, are difficult to predict due to dependency on other factors such as soil moisture, organic matter quality, microbial immobilization and, for phosphorus, chemical immobilization (Nadelhoffer et al., 1992; Robinson & Wookey, 1997). The largest reported increases in net mineralization in Arctic soils in response to experimental warming are approx. 10-fold (Nadelhoffer et al., 1991; Schmidt et al., 1999), but these and most other studies also give much smaller increases (Jonasson et al., 1993; Robinson et al., 1995).
Few experiments address the impacts of increased nutrient availability on arctic ecosystems at nutrient supply rates which are realistic in view of these actual, and potential, rates of deposition and mineralization. For example, the lowest rate of nitrogen addition is commonly 50 kg N ha−1 yr−1, with higher treatments of up to 250 kg N ha−1 yr−1. (Chapin & Shaver, 1989; Potter et al., 1995; Press et al., 1998; Robinson et al., 1998; Shaver et al., 1998). Nutrients have been supplied in NPK fertilizer, so that effects of the individual nutrients, and interactions between them, cannot be distinguished. Given that phosphorus can be limiting within Arctic systems, the effects of increased nitrogen availability are likely to be dependent on the availability of phosphorus.
Despite their limitations, Arctic fertilization experiments have clearly demonstrated that increased nutrient availability affects vegetation community composition. Different vegetation responses have been reported from the tussock tundra of Alaska and subarctic heath of northern Scandinavia, suggested to be due to differing effects of the fertilizer treatments on soil nutrient pools (Chapin et al., 1995; Jonasson et al., 1999). However, in general there is an increase in above-ground biomass of at least some functional groups and, in some cases, decreases in others, including bryophytes, due to altered competitive balance (Jonasson, 1992).
Bryophytes attain their maximum relative importance in terms of biomass and production in tundra vegetation (Richardson, 1981), accounting for up to 70% of the vegetation cover. However, there is little information available on the effects of nitrogen deposition on arctic bryophytes (Bobbink et al., 1998), and the information which does exist does not enable the estimation of a critical load (Woodin, 1997). Bryophytes are efficient scavengers of atmospheric sources of nitrogen (Woodin & Lee, 1987; Jonsdottir et al., 1995) and thus, in the tundra, are not considered to be nitrogen limited (Russell, 1990). Therefore, any increase in nitrogen deposition might be considered to provide an excess supply to arctic bryophytes. Whilst bryophytes rely largely on atmospheric supply of nitrogen, some species may be more dependent on the substratum for phosphorus. The accessibility to bryophytes of any increase in available soil phosphorus due to climatic warming will be dependent on the growth form of the individual species (Bates, 1992).
The objective of this study was to determine the effects on the plant community, in particular the bryophyte component, of nitrogen and phosphorus addition to Arctic heath. Factorial combinations of nitrogen (0, 10, 50 kg N ha−1 yr−1) and phosphorus (0, 5 kg N ha−1 yr−1) were applied for 8 yr to heath communities at Ny-Ålesund, Svalbard. This experiment thus offers a unique opportunity to investigate the effects of long-term, realistic nutrient enhancement. Specifically, the study asks the following. Do the nutrient inputs cause changes in the species composition of arctic heath, in particular that of the bryophyte community? Do the bryophytes become nitrogen saturated, as indicated by their nitrate assimilation capacity? Is the estimated critical load of nitrogen for arctic heath appropriate? Does phosphorus availability influence vegetation responses to nitrogen? and Is there any indication of ‘recovery’ from enhanced inputs of nitrogen and phosphorus within 5 yr?
Materials and Methods
Experimental design and treatments
An experimental site located in mixed tundra heath 1.5 km SE of Ny-Ålesund, Svalbard (78°56′ N, 11°58′ E) was established in 1991. Plots (1.5 × 1.5 m) were located in each of three tundra heath vegetation types, each dominated by one of the dwarf shrubs; Dryas octopetala L., Salix polaris Wahlenb. or Cassiope tetragona (L.) D. Don. Full experimental details are given in Baddeley et al. (1994). Dryas plots received nutrient treatments every year from 1991 to 1998, Salix plots from 1991 to 1997 and Cassiope plots from 1991 to 1993 only. Treatments were factorial combinations of 0, 10 (‘low N’) and 50 (‘high N’) kg N ha−1 yr−1 (as NH4NO3 solution) and 0 and 5 kg P ha−1 yr−1 (as KH2PO4 solution), together with unwatered controls. There were five replicates of each treatment. Treatments were watered on, using a watering can, in four or five applications each summer, each application being equivalent to 2 mm precipitation. Treatments were applied approximately fortnightly from mid-late June (once the whole site was clear of snow) through to early mid August (peak biomass, just before senescence), and thus encompassed almost the whole of the short growing season.
Due to the chemical form of the phosphorus treatment, it also supplies potassium to the vegetation at a rate of 6.3 kg ha−1 yr−1, but the effects of potassium are not investigated separately. Potassium is less limiting in arctic tundra than nitrogen or phosphorus due to its atmospheric input, high mobility of foliar potassium by translocation and leaching, and high availability in the soil (Malmer & Nihlgård, 1980; Jonasson, 1983). Also, the amount of potassium supplied in the experiment is proportionately smaller in relation to both vegetation demand and natural supply than is the amount of phosphorus or nitrogen. It is thus considered unlikely that the potassium input would have a la rge effect on the vegetation.
Bryophyte cover and condition after snowmelt
Within 1 wk of the snow clearing from the plots in 1998, dramatic differences in the colour of the bryophytes between nutrient enriched and control plots were apparent. Thus, in the last wk of June, an assessment was made of the proportion of bryophyte shoots which were green, and thus physiologically active, or brown and presumed inactive. The bryophyte shoot occurring at each intersect of a 20-cm grid within a 1.5-m quadrat was categorised as green or brown. This was undertaken in all three vegetation types.
Ground cover was assessed at the end of July. All touches were recorded at each intersect of a 20-cm grid point quadrat. Data were used to calculate the percentage cover of individual dominant species and of main functional groups; plant litter, bare ground, bryophytes, lichens, shrubs, monocots and herbs. In addition, a species list was made for each plot.
Dicranum scoparium Hedw. and Polytrichum juniperinum Hedw. were selected for more detailed physiological study. These species are ubiquitous and abundant around Ny-Ålesund, Svalbard, and thus were selected as characteristic arctic bryophytes. In addition they exhibit contrasting growth forms.
D. scoparium forms loose tufts on many substrates within numerous habitats worldwide (Smith, 1978). Dicranum leaves are unprotected by any cuticle and it is prone to, but tolerant of, desiccation. Its main source of nitrogen is likely to be from atmospheric inputs.
P. juniperinum has a cosmopolitan distribution and is usually found on well drained acidic soil on heath, moorland, rocks and walls (Smith, 1978). Polytrichum species’ tend to be less prone to desiccation as their leaves have a waxy cuticle and they can change their leaf arrangement to reduce evaporation (Bayfield, 1973). There is an underground stem system with rhizoids which has been shown to transport and store assimilates and water (Collins & Oechel, 1974; Callaghan et al., 1978; Thomas et al., 1988). Thus, for Polytrichum sp., acquisition of nutrients from the substrate may be as, or more, important as that from atmospheric inputs.
D. scoparium and P. juniperinum samples collected from each Dryas plot during the first week in August were dried at 70°C for 48 h. Ground 100 mg subsamples were digested using hydrogen peroxide/sulphuric acid digestion (Allen, 1989). The resulting digests were analysed for total nitrogen on a Technicon autoanalyser (Sterilin Instruments, UK.) using the indophenol-blue reaction, and for phosphate, using the molybdenum-blue reaction (Allen, 1989). Appropriate matrix matched standards were used.
Nitrate reductase activity
On 1 August 1998 samples of the two bryophyte species were collected from each Dryas plot, placed in deionized water and left in natural light overnight to ensure equal water status before the assay. Approx. 100–150 mg of plant material was used (10–15 × 7 mm shoot tips of D. scoparium, 6 P. juniperinum shoots) per sample. To measure induced and constitutive enzyme activity two sets of bryophyte samples were used. Nitrate reductase activity was induced in one set by adding 5 ml of 1 mM KNO3 to samples before assay. Induction time curves for each species showed D. scoparium to reach maximal induced enzyme activity after 3 h and P. juniperinum after 6 h (data not shown). To measure constitutive enzyme activity a second sample set received deionized water for the same induction period. After this period samples were vacuum infiltrated with 2.5 ml of 100 mM potassium phosphate buffer containing 75 mM potassium nitrate (and 0.75% propan-1-ol) and incubated in the dark at 25°C for 1 h. Samples were then placed in a boiling bath for 20 min, cooled and a 1-ml sample removed from each assay for nitrite estimation (Stewart, 1993).
Dried samples collected during the first week in August were also used for spectrophotometric measurement of sucrose (Farrar, 1993).
Data were analysed by ANOVA in Minitab, after checking for normality. First, watered and unwatered controls were compared by one way ANOVA to check for the effects of watering (error df, 8). There was no significant effect of watering on any of the parameters measured except for D. scoparium phosphorus content. This was followed by two factor ANOVA to investigate the effects of nitrogen and phosphorus (error df, 24). For this two factor test ‘control’ plots were those that had received a watering treatment. Any significant main effects or interactions were clarified using the Tukey test. For determination of the critical load of nitrogen it is important to distinguish those effects of nitrogen which are independent of phosphorus supply. This is the case when the interaction term in the two factor analysis is not significant, or the interaction term is significant, but the Tukey test on the interaction means indicates a particular nitrogen-only treatment to differ significantly from the control.
Detrended correspondance analysis, DECORANA, was applied to the species cover abundance data from the Dryas plots using the ordination package PC-ORD (McCune & Mefford, 1997). Correlations were investigated between axis scores and D. scoparium and P. juniperinum tissue nitrogen and phosphorus concentrations (as measures of nitrogen and phosphorus availability).
Bryophyte cover and condition after snowmelt
The importance of bryophytes in the Arctic heath vegetation is demonstrated by observations of the control plots early in the growing season (1 wk after snowmelt). Bryophytes accounted for 43% cover in the Dryas heath, 53% in the Salix heath and 69% in the Cassiope heath. Cassiope control plots contained the highest proportion of green, and therefore physiologically active, shoots, followed by the Dryas and then the Salix plots (35, 27 and 15%, respectively).
After 8 yr, increased nitrogen supply had no significant effect on total bryophyte cover in any of the vegetation types (Fig. 1a–c). However, the low N treatment resulted in a higher proportion of bryophyte shoots being green in Dryas heath (Fig. 1d). There are similar, trends of increased ‘greenness’ with increased nitrogen supply in Cassiope (P = 0.091) and Salix (P = 0.075) plots (Fig. 1e,f), so that when the data from the three heath types are combined there is a significant increase, by almost half, in the proportion of bryophyte shoots which are green in low N plots, with no further increase in response to high N (Table 1).
Table 1. xF-values and probabilities from ANOVA for the measured parameters
N × P F2,24
N × P
(kg N ha−1 yr−1)
Factors used in two way ANOVA were nitrogen (N) and phosphorus (P). ns, not significant. Differences between N treatment main effect means were clarified using the Tukey test and are denoted by different small case letters (for direction of response see figures). Tukey tests for significant interaction terms are not shown, but are mentioned in the text where relevant.1For all three vegetation types combined error df, 82.2The percentage cover of graminoids and herbs was too small to fulfil the assumptions for ANOVA.3NRA, nitrate reductase activity.
In contrast to nitrogen, addition of phosphorus increased bryophyte cover by 13, 16 and 15% in Dryas, Salix and Cassiope plots, respectively (Fig. 1a–c). Also, the proportion of bryo–phyte shoots that were green was increased by up to a third in all three vegetation types (Fig. 1d–f). There was no additional effect caused by the application of both nutrients together.
Species composition of Dryas plots
At the height of the growing season, total live vegetation cover in Dryas plots was not significantly affected by increased nitrogen supply, but high N increased litter cover by 10% (a two thirds increase in litter) (Fig. 2). Conversely, increased phosphorus supply increased total vegetation cover by 11% and reduced the ground cover of litter (Fig. 2).
Lichen cover was strongly affected by nutrient addition (Fig. 2). The low N treatment had little effect, but high N reduced abundance by almost half. The combination of either nitrogen supply rate with phosphorus exacerbated this negative response, reducing cover by 60% when compared with control plots.
Total bryophyte cover was not significantly affected by increased nitrogen supply, but addition of phosphorus almost doubled it (Fig. 2). Further analysis was carried out for the dominant bryophyte species. P. juniperinum showed an increase in abundance in response to either nitrogen or phosphorus alone, and a greater than additive response to the combination of these nutrients (Fig. 3a). D. scoparium cover was more variable, such that responses were not significant (although the pattern was of a negative response to fertilization, Fig. 3b). Pohlia wahlengbergii (Web & Mohr) Andrews. showed no overall response to nitrogen, but an almost 10-fold overall increase in response to phosphorus (Fig. 3c). Cover of Schistidium sp. was increased by either high N or phosphorus alone, but was decreased by the combination of both nutrients (Fig. 3d).
There were no significant effects of either nutrient alone on total cover of shrubs, graminoids or herbs (Fig. 2), although for the latter two groups cover was so small that detection of any significant effect would be unlikely (and so they are combined in Fig. 2).
Total species number was reduced by increasing nitrogen supply alone, with two species fewer in high N treated vegetation (Fig. 4). Whilst phosphorus alone had little effect, the addition of phosphorus to vegetation also receiving low N caused species number to increase by 2.6 compared with unfertilized vegetation. The combination of high N plus phosphorus caused no such increase (Fig. 4).
Ordination (DECORANA) showed distinct separation on Axis 1 of those plots that had received both nitrogen and phosphorus from all the other plots (Fig. 5). Axis 1 accounted for 29% of the variability in the data. There was less distinct separation on Axis 2 (which accounted for 14% of the variance), but again the mean axis score for combined nutrient treatment plots was higher than that for all other plots. Tissue phosphorus concentrations of both D. scoparium and P. juniperinum were closely correlated to Axis 1, as was, to a lesser extent, the tissue nitrogen concentration of D. scoparium (Table 2). Tissue phosphorus concentration of D. scoparium was also reasonably well correlated with Axis 2.
Table 2. Correlations between foliar N and P concentrations of Dicranum scoparium and Polytrichum juniperinum and the Dryas octopetala heath plot DECORANA axis scores
ns, not significant.
In both species, nitrogen addition increased tissue nitrogen concentration, which more than doubled in the high N treatment (Fig. 6). Considering the bryophytes receiving nitrogen only, tissue nitrogen increased linearly with nitrogen in P. juniperinum (y = 0.107x + 4.47, R2 = 92%, P < 0.001), but tended to saturation at high N in D. scoparium (y = 3.82(1 − e−0.1x) + 3.3, R2 = 94.4%, P < 0.001). Phosphorus concentration was increased approx. threefold by phosphorus addition (Fig. 6). Nitrogen addition significantly increased tissue phosphorus content of D. scoparium, possibly due to reduced growth (Fig. 6).
Constitutive nitrate reductase activity was of similar magnitude for both species (Fig. 7). Nitrogen treatment reduced both constitutive and inducible nitrate reductase activity in both species (Fig. 7). In D. scoparium loss of activity was significant in the high N treatment, in P. juniperinum it was significant in both N treatments. Phosphorus treatment often resulted in greatly increased activity of the enzyme (Fig. 7).
Soluble carbohydrate content
Sucrose content of P. juniperinum was increased by both low and high nitrogen supply. That of D. scoparium was increased only in response to the combination of nitrogen and phosphorus addition (Fig. 8).
Effects of fertilization on the arctic heath community
Nitrogen addition had no significant effect on the cover of live vegetation in the Dryas heath, but increased the amount of plant litter. In subarctic heath, fertilization with NPK (100 kg ha−1 yr−1 of N and P) also increased dead standing biomass and litter, and, as the total standing biomass was unaltered, this was interpreted as faster turnover of plant material (Press et al., 1998). In the high Arctic, at Ny-Ålesund, a different mechanism gave rise to increased litter; during winter 1993–4 an unusually warm wet autumn delayed hardening and caused increased winter damage, resulting in increased litter in NPK fertilized plots (50 kg ha−1 yr−1 of N and P) (Robinson et al., 1998). Thus, it is likely that damage caused by the same warm spell is responsible for the increased litter cover in our high N plots. By contrast, phosphorus addition to our plots increased the cover of live vegetation; presumably a result of existing individuals becoming larger and/or recruitment increasing during the 8 yr experiment. The increase in species number in vegetation receiving phosphorus and nitrogen suggests that recruitment might have at least contributed to the increased cover. Seedlings were not actually observed in our plots, but ingress of new individuals into NPK fertilized plots also at Ny-Ålesund was observed by Robinson et al. (1998).
Bryophyte and lichen responses
The only plant functional types to show significant responses to fertilization were the bryophytes and lichens, which were expected to be sensitive due to their low nutrient requirement and ability to assimilate deposited nutrients over their entire surface area. Surprisingly, bryophyte cover was unaffected by increased nitrogen supply, although it appears that this is a net result of individual species showing different responses (see the Bryophytes: a valid functional type? section). The effect of phosphorus was dramatic with bryophytes further increasing their dominance of ground cover.
In subarctic Sweden total bryophyte cover was reduced by 50% in response to the addition of 100 kg ha−1 yr−1 of N and P to a shrub heath (Potter et al., 1995). Similarly in Alaskan tussock tundra combined nutrients reduced the growth of Aulocomnium sp. (Chapin & Shaver, 1985). However, evidence from the high Arctic provided by this study and colleagues (Klokk & Ronning, 1987; Robinson et al., 1998), and from a subarctic fellfield (Jonasson et al., 1999), show the opposite, with fertilization increasing bryophyte abundance. Thus, in closed vegetation, increased shrub and graminoid growth in response to fertilization results in most bryophytes being shaded out (Press et al., 1998; Jonasson et al., 1999). In open vegetation where bryophytes form a dominant component of the vegetation, as is typical of the tundra, their response is not mediated by that of higher plants and their abundance is increased.
Lichens showed a significant response to nitrogen, suffering a considerable loss of cover when nitrogen was added, and even further loss with the combination of nutrients. There have been many studies of relationships between lichens and air pollutants (Nash & Gries, 1995), but few deal with nitrogenous pollutants. Thallus nitrogen content has been found to be proportional to nitrogen deposition rates (Bruteig, 1993; Hyvarinen & Crittenden, 1998), but there have been few manipulative studies of effects. The nitrogen content and biomass of Cladina rangiferina sprayed with nitric acid over a 2-yr period was stimulated, whilst Peltigera apthosa was unresponsive (Scott et al., 1989; Hallingback & Kellner, 1992). Conversely, addition of NPK (100 kg ha−1 yr−1 of N and P) to subarctic heath caused a 75% reduction in lichen abundance, with all four dominant species recorded suffering loss of cover (Press et al., 1998). Thus, although some lichen species may benefit from increased nutrient supply, the majority suffer a decline. It may be that nitrogen, and possibly phosphorus, are directly damaging. In Arctic regions, fertilized lichens might also be preferentially grazed by reindeer (Rangifer tarandus). Lichens can account for up to 60–70% of their winter food (Longton, 1997), and our experimental site is grazed.
Bryophytes: a valid functional type?
Increased nutrient supply altered the abundance of individual bryophyte species, with some increasing in abundance whilst others declined. The bryophyte flora was becoming dominated by what may be categorised nutrient–philic species, in particular, Polytrichum juniperinum and Pohlia wahlengbergii, and cover of other dominant species, Dicranum scoparium and Schistidium sp., was reduced. Individualistic responses of higher plants to greater rates of nutrient addition in the subarctic have been well demonstrated (Shaver & Chapin, 1980; Press et al., 1998). The subarctic fertilization experiment of Potter et al. (1995) (100 kg ha−1 yr−1 of N and P) also demonstrated species specific bryophyte responses, with the biomass and relative importance of Polytrichum commune increased and that of Hylocomium splendens decreased.
Grouping of species into functional types, as advocated by Chapin et al. (1996), is useful for extrapolative and predictive purposes, but disguises such individualistic species responses. Bryophytes tend to be lumped as a single group in functional type studies. Interestingly Chapin et al. (1996) subdivide bryophytes, on the basis of their peat forming capacity, into Sphagnum and non-Sphagnum species. However, their data, presented in a cluster diagram, do not reflect this choice well as the major division within the bryophytes as presented occurs between Polytrichum and other bryophyte species (including Sphagnum). This division reflects the relative structural complexity of Polytrichum species with their greater potential for nutrient and water uptake from the soil. Thus, on the basis of their data, ours and those of Potter et al. (1995), more appropriate functional divisions appear to occur between Polytrichum species, Sphagnum species and other bryophytes. As bryophytes exhibit such dominance in the Arctic, and have such important ecological functions, it is appropriate that they are not treated as one group but that such divisions are made.
Co-limitation by N and P
The importance of both phosphorus and nitrogen in influencing the vegetation has been clearly demonstrated. The interactive effects of the two nutrients, probably best illustrated in the changes in cover of P. juniperinum, suggest that they may actually be co-limiting. This is supported by the ordination of the species cover data for the Dryas heath plots which shows distinct separation of those plots receiving both nitrogen and phosphorus. The availability of both nutrients, as indicated by bryophyte tissue contents, was strongly related to the main axis of variation. Species number was also increased only by the combination of low N and phosphorus. Together these data demonstrate that co-limitation by nitrogen and phosphorus occurs in this high Arctic heath community. The response to atmospheric nitrogen deposition will, therefore, depend on the phosphorus status of the community. Similarly, responses to any changes in net mineralization due to soil warming will be dependent upon the ratio in which nitrogen and phosphorus become more available to the vegetation.
Sensitivity of bryophyte physiological responses to fertilization
Both phosphorus and, to a lesser extent nitrogen, increased the proportion of green bryophyte shoots, thus apparently increasing potential total bryophyte productivity. The increased ‘greenness’ of plots was visible throughout the summer, but particularly so early in the season, suggesting that the nutrients effectively extend the duration of the growing season for the bryophytes. For phosphorus the clear outcome is increased bryophyte cover. The lack of increase in cover in response to nitrogen, despite the apparent physiological stimulation, is perhaps surprising, but explanation might be provided by study of the growth of individual species and individual shoots.
The sucrose content of P. juniperinum does provide firmer evidence for increased carbon assimilation in response to nitrogen fertilization. Sucrose content reflects a combination of both carbon assimilation rate and growth rate (sink strength). To achieve the observed increases in both cover and sucrose content, the net carbon assimilation of the P. juniperinum population must have increased. For D. scoparium, the slight increase in sucrose content in response to combined fertilizer treatment may be, at least in part, a result of reduced sink strength, since cover also tended to decrease.
Whatever the mechanism of effect, a dramatic finding was the persistence, 5 yr after nutrient additions had ceased, of increased ‘greenness’ of the bryophyte cover in the fertilized Cassiope heath plots. Effects were as large as those seen in the Dryas and Salix heaths, which had received annual nutrient additions in each of the 7 yr before sampling. This suggests that the added nutrients are still being tightly held within the bryophyte layer of the Cassiope heath and implies that any reversal of effect will be slow. This lack of ‘recovery’ clearly demonstrates the potential for long-term ecological change caused by even small amounts of nitrogen deposition, due to the high nutrient conservatism of Arctic ecosystems.
As expected, the tissue nutrient content of both D. scoparium and P. juniperinum increased with increasing experimental supply, demonstrating their close coupling with atmospheric inputs. The linear relationship between nitrogen input and tissue nitrogen concentration in P. juniperinum suggests its usefulness as a bioindicator of deposition.
Although the tissue nitrogen data suggest continued assimilation of nitrogen with increasing input, nitrate reductase activity in P. juniperinum was inhibited by both the low and high N treatments, whilst that in D. scoparium was reduced by the high N treatment only. This enzyme is end product inhibited (Woodin et al., 1985), the implication being that uptake of nitrogen from atmospheric sources is decreased as the bryophyte becomes saturated. The strong reduction in inducible nitrate reductase activity in P. juniperinum suggests physiological nitrogen saturation at just 10 kg N ha−1 yr−1. Nitrate reductase activity might show earlier signs of saturation in P. juniperinum than in D. scoparium due to its greater morphological differentiation. It has a cuticle and rhizoids, making it less prone to leaching and allowing greater opportunity for nitrogen storage and remobilization, lessening the requirement for assimilation. Thus, although different species have different sensitivities to increased deposition, both species tested showed evidence of nitrogen saturation in reduced dependence on the scavenging tactics provided by nitrate reductase induciblity.
Bryophytes have been demonstrated to provide efficient buffering of nitrogen inputs to vegetation by scavenging and retaining all NH4 and NO3 deposited (Woodin & Lee, 1987; Jonsdottir et al., 1995). If however, the bryophytes become saturated, inorganic nitrogen will pass through the bryophyte layer, becoming available for soil microbes and higher plants, and also potentially being lost from the system through leaching and/or N2O emission. Thus, the nitrate reductase data suggest that after several years application of 50 kg N ha−1 yr−1 to the tundra heath, the bryophytes will no longer be scavenging all the nitrogen deposited; the system is less buffered by their presence. This ties in with a previous finding that, following an application of 15NH415NO3 to the experimental Salix heath plots, 15N recovery from the soil and vegetation of previously unfertilized plots was significantly < that from plots which had received the high N treatment (Alexander et al. unpublished), suggesting that nitrogen loss was occurring as a result of saturation.
The very significant influence of phosphorus availability on nitrate reductase activity is likely to result from the alleviation of phosphorus limitation increasing bryophyte nitrogen demand. Phosphorus stimulates photosynthesis and export of photosynthetic products from the chloroplast to the cytosol, some of which stimulate expression and activation of nitrate reductase (Kaiser & Brendle-Behnisch, 1991; Lambers et al., 1998). This again clearly demonstrates that the response of bryophytes to nitrogen inputs will be highly dependent on their phosphorus status.
Given the important functional roles of bryophytes in Arctic systems (see the Validation of the critical load of nitrogen for Arctic heath section), the changes in their abundance, tissue nutrient status and nitrate assimilation capacity caused by increased nitrogen and phosphorus supply are likely to have significant consequences for nutrient cycling within the whole system.
Validation of the critical load of nitrogen for Arctic heath
The generally accepted working definition of the critical load for total nitrogen deposition is ‘A quantitative estimate of an exposure to deposition of N as NHx and/or NOy below which empirically detectable changes in ecosystem structure and function do not occur according to present knowledge’ (Hornung et al., 1995). This study of an Arctic heath has concentrated on bryophytes which form the dominant plant cover, stabilizing the ground surface. They have a strong influence on soil temperature and moisture due to their high water-holding and thermal insulating capacities. They also affect ecosystem nutrient cycling through their very efficient nutrient scavenging and retention, and slow decomposition (Longton, 1997). Thus, any change in their abundance or nutrient dynamics will have major effects on ecosystem structure and, more particularly, function.
As expected, 50 kg N ha−1 yr−1 was clearly demonstrated to be above the critical load for tundra heath, reducing the nitrate scavenging capacity of the bryophytes, which may have affected nutrient retention by the whole system. This rate of supply was apparently supra-optimal for ‘greenness’ of the total bryophyte cover and for lichen cover (Tukey on interaction means, P < 0.05), the latter of which has potential consequences for ground surface stability and for the nitrogen budget of the system. An increase in litter was apparently due to higher plants being rendered more susceptible to winter damage by this rate of nitrogen input.
Most importantly, evidence is also provided that even 10 kg N ha−1 yr−1 is causing community change. It increased the proportion of green bryophyte shoots (in Dryas heath and in all three heath types combined), and reduced the nitrate scavenging ability of P. juniperinum, with possible consequences for nitrogen retention in the system. Therefore, just 10 kg N ha−1 yr−1 is causing some change in the structure and function of the community and so is in excess of the critical load. This provides first time validation that the critical load for Arctic heath lies at the lower end of a previous estimate of 5–15 kg N ha−1 yr−1 (Hornung et al., 1995).
Nitrogen also had effects which were dependent on phosphorus supply. With added phosphorus, low N decreased lichen cover and both rates of nitrogen supply caused marked increase in P. juniperinum cover and measurable change in total vegetation composition. Thus, although the rate of phosphorus addition employed in this experiment is probably rather high with respect to any change which may occur as a result of climate change, these findings do demonstrate that the critical load for nitrogen will decrease with either spatial or temporal increase in phosphorus availability to Arctic heath vegetation.
Finally, it is clearly demonstrated that recovery from exceedence of the critical load for nitrogen within this ecosystem will be slow.
We are indebted to the NERC (GR9/3433), the CEC TMR Programme and the Scottish International Education Trust for the funding which enabled this research. We are especially indebted to Nick Cox and the NERC for the use of Harland Huset at Ny-Ålesund.